amazon-scraping

A Python scraper for Amazon product data: name, price, stock, rating and image from product pages, every color/size/model variation of a product with its own price, and all seller offers with condition, seller rating and fulfillment. Output as Python dicts, JSON or CSV.

pip install amazon-scraping

Why a plain request fails

Amazon sits behind AWS WAF. Every request is scored on its IP, headers and recent activity, and requests or a headless browser on a datacenter IP gets a CAPTCHA page or a block within a few calls. Building around it means maintaining a pool of residential proxies, browser-like headers and fingerprints.

This package fetches pages through the Scrape.do API instead:

• Product pages go through the Amazon plugin (/plugin/amazon/?url=...&geocode=us&zipcode=10001&output=html), which returns the full page with US pricing for the ZIP code you set. No render=true, no super=true: the plain call came back 200 in under 4 seconds in October 2026, and it is cheaper.
• Seller offers come from Amazon's own aodAjaxMain panel through the standard API.
• Structured JSON, if you'd rather skip HTML: pdp() and offer_listing() call the Amazon Scraper API endpoints.

You need a Scrape.do API token; the free plan includes 1,000 credits a month.

Quick start

from amazon_scraping import AmazonScraper

scraper = AmazonScraper(token="YOUR_SCRAPEDO_TOKEN")  # or set SCRAPEDO_TOKEN; zipcode="10001" by default

product = scraper.get_product("B0BLRJ4R8F")          # an ASIN or a product URL
print(product["name"], product["price"], product["rating"], product["in_stock"])

variations = scraper.get_variations("https://www.amazon.com/Apple-iPhone-Plus-128GB-Blue/dp/B0CG84XR6N/")
for v in variations:
    print(v["asin"], v.get("color"), v.get("size"), v["price"])

offers = scraper.get_offers("B0BN969PGL")
for o in offers:
    print(o["seller"], o["price"], o["condition"], o["seller_rating"], o["ships_from"])

Command line

export SCRAPEDO_TOKEN=your_token

# products, one ASIN or URL per line, CSV
amazon-scraping product --input asins.txt --format csv --output products.csv

# every variation of one product
amazon-scraping variations "https://www.amazon.com/dp/B0CG84XR6N/" -f csv -o amazon_variations.csv

# all seller offers
amazon-scraping offers B0BN969PGL -f csv -o amazon_seller_offers.csv

# structured JSON from the Amazon Scraper API
amazon-scraping pdp B0BLRJ4R8F

Output

Product (from the fictional test page in tests/fixtures):

{
  "asin": "B0BLRJ4R8F",
  "name": "Example Basics Portable Laptop Stand, Height Adjustable, Silver",
  "price": 1029.87,
  "currency": "USD",
  "in_stock": true,
  "rating": 4.6,
  "image": "https://m.media-amazon.com/images/I/example-stand.jpg",
  "url": "https://www.amazon.com/Some-Stand/dp/B0BLRJ4R8F/"
}

Variation rows carry one column per dimension the product has (color, size, style, service_provider, ...), so the CSV always matches the product's own variation model.

How it works

Product page. Title from #productTitle, price from .a-price-whole + .a-price-fraction (turned into a float, so 1,029. + 87 becomes 1029.87), rating from .AverageCustomerReviews, image from img#landingImage. If #outOfStockBuyBox_feature_div is on the page the product is out of stock and the price is None instead of a stale or missing value.

Variations. Amazon puts each variation dimension in a div#inline-twister-row-<dimension>; every option is an li with its own data-asin. The scraper reads the dimensions on the base page, sorts them (color, size, style, ...), and walks the option tree depth first, loading each option's own page, because the available options change with the selection (some sizes only exist in some colors). It reads the selected options and the price from each final page and never visits an ASIN twice. Budget one request per variation page; max_pages caps it. Note that html.parser lowercases attribute names, so options marked data-initiallyUnavailable="true" are matched in lowercase and skipped.

Seller offers. The "Other sellers" panel loads from /gp/product/ajax/aodAjaxMain/?asin=<ASIN>. Each offer has a div#aod-offer-soldBy; the scraper climbs to the container holding the price, reads the seller link (no link means Amazon sells it), seller rating, rating count, positive feedback percentage, ships-from and condition, and skips the sticky copy of the Buy Box offer inside #aod-sticky-pinned-offer so it isn't counted twice. One offer, or none, is a valid result: the count depends on the ASIN and the delivery location.

Limits

• Reviews are not scraped. Since May 2026 the public /product-reviews/ pages return Page Not Found to logged-out clients and review text was removed from the product page HTML. Login-protected content is off-limits; use the Amazon Scraper API PDP endpoint for featured reviews.
• Markup changes. Amazon changes its HTML. Parsers raise ValueError on pages without a product title (usually a CAPTCHA page) and return None/empty values for missing fields rather than guessing.
• Errors. Non-200 responses raise ScrapeError with the status code. Server errors and 429s are retried twice; client errors are not.

Responsible use

Stick to public pages (products, prices, offers), keep request rates reasonable, and stay away from anything behind a login. Amazon's Conditions of Use restrict automated access; you are responsible for how you use the data and for the laws that apply to you. This is not legal advice.

Development

git clone https://github.com/bugrahansss/amazon-scraping
cd amazon-scraping
pip install -e .
python -m unittest discover -s tests -v

The tests run offline against fictional fixture pages.

License

MIT

Source: Complete 2026 Amazon Scraping Guide: Product Data, Prices, Sellers, and More